Jian Wang, Haq Amin Ul, Afzal Noman, Khan Shakir, Alsolai Hadeel, Alanazi Sultan M, Zamani Abu Taha
School of Artificial Intelligence, Neijiang Normal University of Sichuan, Neijiang, Sichuan, 641100, China.
Institute of Telecommunications, Computer Science and Photonics, Scoula Superiore Sant'Anna (SSSA), Pisa, Via Moruzzi 1, 56124, Italy.
Sci Rep. 2025 Jul 2;15(1):22945. doi: 10.1038/s41598-025-03960-2.
Accurate Lung cancer (LC) identification is a big medical problem in the AI-based healthcare systems. Various deep learning-based methods have been proposed for Lung cancer diagnosis. In this study, we proposed a Deep learning techniques-based integrated model (CNN-GRU) for Lung cancer detection. In the proposed model development Convolutional neural networks (CNNs), and gated recurrent units (GRU) models are integrated to design an intelligent model for lung cancer detection. The CNN model extracts spatial features from lung CT images through convolutional and pooling layers. The extracted features from data are embedded in the GRUs model for the final prediction of LC. The model (CNN-GRU) was validated using LC data using the holdout validation technique. Data augmentation techniques such as rotation, and brightness were used to enlarge the data set size for effective training of the model. The optimization techniques Stochastic Gradient Descent(SGD) and Adaptive Moment Estimation(ADAM) were applied during model training for model training parameters optimization. Additionally, evaluation metrics were used to test the model performance. The experimental results of the model presented that the model achieved 99.77% accuracy as compared to previous models. The (CNN-GRU) model is recommended for accurate LC detection in AI-based healthcare systems due to its improved diagnosis accuracy.
在基于人工智能的医疗系统中，准确识别肺癌是一个重大医学问题。人们已经提出了各种基于深度学习的方法用于肺癌诊断。在本研究中，我们提出了一种基于深度学习技术的集成模型（CNN-GRU）用于肺癌检测。在所提出的模型开发中，卷积神经网络（CNN）和门控循环单元（GRU）模型被集成起来，以设计一个用于肺癌检测的智能模型。CNN模型通过卷积层和池化层从肺部CT图像中提取空间特征。从数据中提取的特征被嵌入到GRU模型中，用于肺癌的最终预测。使用留出验证技术，利用肺癌数据对模型（CNN-GRU）进行验证。采用旋转和亮度等数据增强技术来扩大数据集规模，以有效地训练模型。在模型训练期间应用随机梯度下降（SGD）和自适应矩估计（ADAM）等优化技术来优化模型训练参数。此外，使用评估指标来测试模型性能。该模型的实验结果表明，与先前的模型相比，该模型的准确率达到了99.77%。由于其提高的诊断准确率，（CNN-GRU）模型被推荐用于基于人工智能的医疗系统中进行准确的肺癌检测。
